Guidewire (GW) tracking in a collateral channel (CC) is an important step during retrograde chronic total occlusion (CTO) percutaneous coronary intervention (PCI). The aim of this study was to create a prediction score model for CC GW crossing success.We analyzed data on 886 CCs included in the Japanese CTO-PCI Expert Registry during 2016. CCs were categorized as septal (n=610) and non-septal (n=276). CCs were randomly assigned to derivation and validation sets in a 2:1 ratio. The score was developed by multivariate analysis with angiographic findings. Small vessel, reverse bend, and continuous bends were independent predictors in the septal CC subset. Small vessel, reverse bend, and corkscrew were independent predictors in the non-septal CC subset. The extent of intervention was easy, intermediate, and difficult at 92.9%, 57.4%, and 16.7% in the septal CC subset and 91.7%, 54.3%, and 19.0% in the non-septal CC, respectively, in the validation set. The area under the receiver-operator characteristics curve was >0.7 in derivation and validation sets of both CC subsets.The prediction score model can suggest grading of the difficulty of CC GW crossing based on angiographic findings for each type of CC.